Why Do We Want Low Voltage?

M

mercury529

Limp Gawd
Joined
Dec 19, 2007
Messages
173
Hello,

I am fairly new to overclocking. I had a question about voltage. Everyone seems very concerned with keeping a processor's voltage as low as possible.

1.) What are the reasons for wanting to keep the voltage low?

I have a Q6700 and am speaking specifically about voltages beneath 1.45v. Is the only incentive for wanting to keep the voltage low to decrease power consumption/heat? If so, I have a few questions.

I found this equation in the overclocking guide: P=CFV^2. I can reliably overclock my processor at 400 FSB and a multiplier of 8 at a voltage of 1.33750. Assuming the equation provided is true, that means my power consumption should be the same if I overclock at 333 FSB with a multiplier of 10 at a voltage of 1.465. Both settings allow me to run my ram at 800 MHZ (DDR2-800).

2.) Is the equation correct? Is it all that needs to be taken into account for processor power consumption?
3.) Does power consumption directly correlate to heat production?
4.) Would there be any disadvantage to running at 333 x10 instead of 400x8?

Thanks a lot for any info you can give.
 
Depending on the process used to make the chip (90nm, 65nm, 45nm etc) there is a 'safe' level of voltage that won't be too bad for the chip, but as you push it higher you exponentially reduce the life of a chip.

Excessive voltage will kill a cpu in days.
 
First of all, your interpretation of the equation is wrong. It doesn't depend on the FSB speed, it depends on the CPU speed. The specific FSB and multiplier used don't matter, what determines the power usage is the CPU frequency and the voltage used. Also, the equation isn't completely accurate, since there are other factors that affect power usage as well.

Aside from that, the above poster is spot on. Voltage increases power consumption and heat, and also voltages past a certain level will damage the CPU. With 65nm Core 2 CPUs, the threshold is around 1.5v, so I highly suggest you keep things below that point.
 
Also this is why we look for low VIDs on CPUs, because if a CPU is binned as lower voltage, it is more likely (but not guaranteed - it's overclocking, nothing is) that it will overclock higher than a CPU with a higher VID since it needs a lower amount of voltage to reach its stock speed.
 
1.5v on 65nm? I ran 1.575 for over a year. Keep the heat down and you're fine. I'm running 1.45v on my 45nm.
 
Thanks for the clarification that the F stood for CPU Frequency not FSB speed. I read the line above it that stated that heat goes up when the FSB increases and automatically associated that with the variable F. That clears it up for me.

Does raising the voltage (within the 1.25-1.45v range) always reduce the life of a chip? Or does it only decrease the life of the chip only when it results in higher operating temperatures?
 
Raising the voltage always reduces the life of a CPU somewhat. However, this degradation is mostly insignificant until you reach a certain point, which depends on the specific type of CPU. After you pass that point, any increase in voltage will drastically shorten the life of the CPU. An early sign of this is when the CPU ceases to be stable at previously stable overclocks.

Degradation does not depend entirely on heat, although it is a factor. The general rule of thumb is that unless you're using sub-ambient cooling, you should always stay within the limits. This means that if you're using air or water cooling, don't go past 1.5v with 65nm Intel CPUs or 1.4v with 45nm chips. If you're using chilled water, phase change, dry ice, LN2, etc., then it's safe to use high voltages for short periods of time. However, for general usage, it's always a very good idea to stay within the limits.
 
You must log in or register to reply here.
Back
Top